Wayfinding

by M. R. O'Connor

  Before O’Keefe discovered place cells in the early 1970s, neuroscientists knew that the hippocampus was involved in memory, though what kind of memory was disputed. One of Nadel and O’Keefe’s professors at McGill was the neuropsychologist Brenda Milner, the first person to have studied the patient H.M. and written a paper about the nature of his amnesia after the removal of part of his temporal lobe to treat severe epilepsy. Milner recognized that there were different types of memory systems and learning, and that H.M.’s amnesia was episodic in nature. But later theories of the hippocampus and memory argued that the hippocampus was responsible for something else as well: semantic memory, the recollection of facts. Reconciling Milner and these other theories of the hippocampus was one of the major challenges facing O’Keefe and Nadel. Their theory predicted that the hippocampus was the core of a neural system that provided a spatial framework for storing memories of what happened in a particular place, not facts but events. “The idea was that episodes were built upon the basic spatial framework through the addition of a linear sense of time amongst other higher-order cognitive capacities.”

  It wasn’t until the 1990s and the advent of virtual reality—the ability to create computer simulations of large-scale environments—that neuroscientists could use MRI on immobile people to fully understand what parts of the brain were activated during navigation and memory recall—thereby confirming this idea. The earliest virtual reality tests used a first-person shooter game, Duke Nukem, stripped of all of the guns and fighting, leaving only the mazelike environment that subjects could wayfind through. In 2001, O’Keefe and several other researchers at University College London designed a study whereby epileptics who had undergone lobectomies in their right or left temporal lobes were asked to explore a town in the video game environment for about an hour, during which they encountered different characters. They were then tested on their ability to draw maps of the environment and their memory of the events. They found that whereas those with right temporal lobe lobectomies were impaired in navigation and spatial memory, those with left-side lobectomies were impaired in the episodic memory tests, suggesting that the hippocampus was indeed a critical part of the brain for both cognitive mapping and episodic memory.

  In subsequent years, scientists discovered other critical cells in the hippocampus and a remarkable level of plasticity in hippocampal physiology. Some of these other cells include head-direction cells, which discharge in relation to which way our head is pointed on the horizontal plane, and grid cells, which fire as we roam an environment and build a coordinate system for navigating. There’s also evidence that the richness and complexity of an environment influences the quantity of neurons in the hippocampus. In 1997, for instance, three researchers, including Rusty Gage at the Salk Institute, found that mice exploring enriched environments—paper tubes, nesting material, running wheels, and rearrangeable plastic tubes—had forty thousand more neurons than a control group. These additional neurons resulted in an increase in hippocampal volume of 15 percent in the mice and significant improvements in their performance on spatial learning tests. The researchers concluded that a combination of increased numbers of neurons, synapses, vasculature, and dendrites led to the animals’ enhanced performance on the tests.

  Today, an even fuller picture of how hippocampal cells interact and build spatial representations for orientation and navigation has emerged. As Kate Jeffery and Elizabeth Marozzi explained in Current Biology, multiple sensory systems, from vision to touch and olfaction, converge upstream of the hippocampus and are “combined into supra modal representations such as landmarks, compass cues, boundaries, linear speed,” which are then passed on to the place cells. At the same time, head-direction cells give us a sense of direction, firing only when the head is pointed in a particular direction, like a kind of neural compass. Border cells seem to signal the distance and direction from a boundary that could be an obstruction, gap, or step. Grid cells are thought to represent space at different scales by using both environmental and self-motion information to generate information about distance. They fire in an environment in a fascinating pattern: a hexagonal array that extends in all directions, and they are one synapse upstream of place cells. While the interaction among these different cells is still somewhat mysterious and the topic of much research, it’s likely that grid cells send information to the place cells used for path integration while also receiving information in return. Very accurate navigators seem to show more hippocampal activation and engagement, and navigational experience itself seems to increase volume of the brain, as the study of London’s taxi drivers originally showed. The cognitive mapping system is, surprisingly, not dependent on vision. There is evidence that blind people formulate cognitive maps. Blind individuals use kinesthetic and motor signals to dead reckon, and seem to be better at this than sighted individuals.

  Matt Wilson is a neuroscientist at MIT. He calls the classic experiments of running rats through mazes and listening to their brain cells “eavesdropping,” and he has been doing it for years to try and understand what this system of cells has to do with memory. Testing the relationship takes ingenuity. “If you damage the hippocampus in humans or rodents, you will lose the capacity to form memories of life experience. Now, it’s hard to ask those life-experience questions to a rat. But what you can do is test them on another kind of memory: just ask a rat to go back to a place it’s already been. Rats have very good spatial memory.” The connection between spatial navigation and memory of life experiences, according to Wilson, is time. “Both [navigation and memory] depend on a critical function, linking things in time,” he said. “It is how you put the pieces together, how you create an internal narrative of your experience. It’s not a record, or videotape of experience. It involves evaluating, selecting, and sorting things. Rats create an experience of moving around in space. We create the stories of our lives.”

  * * *

  How can we be so sure that space was the primary concern of place cells? What if space just matters more to rats, the favored study subject of tens of thousands of maze experiments conducted since the early twentieth century, and there are other domains of experience the hippocampus is sensitive to? Some neuroscientists believe hippocampal cells are actually implicated in a far greater scope of human cognition than spatial representations, and they doubt that our brains are even really building representations that are structurally analogous to an allocentric map. Maybe the cognitive map is much more flexible than that, and the hippocampus encodes and builds maps for many dimensions of human experience beyond space—everything from time to social relationships, sound frequencies, even music.

  One warm fall day, I walked from the MIT campus, where Edward Tolman had once studied and the amnesiac patient H.M. had spent countless hours being observed, and crossed the Charles River to Boston University to meet one of the most prominent dissenters of the cognitive map theory—Howard Eichenbaum, the director of the Center for Memory and Brain and the Cognitive Neurobiology Laboratory. I climbed the stairs to Eichenbaum’s second-floor office and knocked on the door. I was greeted by a white-haired, mustachioed man from behind a desk covered in stacks of papers, likely related to his work as editor in chief of the scientific journal Hippocampus. Hanging on the wall in a frame behind him, I saw a poem, “The Experiment with a Rat” by Carl Rakosi:

  Every time I nudge that spring

  a bell rings

  and a man walks out of a cage

  assiduous and sharp

  like one of us

  and brings me cheese.

  how did he fall

  into my power?

  He put his feet up on a chair and asked me, “So what is navigation anyway?”

  I laughed. It was the simplest question. But despite thinking about almost nothing else for several years, I had yet to find a simple answer. It is fundamentally the task of getting from one place to another. But it can entail so many different strategies in both animals and humans, so many scales and perspective
s, that it hardly seems to fit under one action, process, or skill. Instead, it involves multiple arrays of cognition and possible problem-solving techniques. Thus far, scientists have created a multitude of categories to try and capture it. Vector navigation involves moving along a constant bearing relative to a cue that could be magnetic, celestial, or environmental. Piloting is defined as navigating relative to familiar landmarks. True navigation generally means wayfinding toward a distant, unseen goal. Dead reckoning, also called path integration, is keeping track of every stage of a journey in order to compute one’s location.

  As it turns out, both rats and humans are the worst at path integration, which is precisely the kind of navigation that cognitive-map theorists propose the hippocampus does. In Eichenbaum’s opinion, this is extremely problematic. “One of my complaints about the path integration theory is how bad we are at it,” he said. Dead reckoning is applicable in short distances at local scales, but it is a strategy that isn’t actually advisable in real-world navigation because it is so prone to accumulated error (except, it would seem, for those who have mastered complex environments like the Arctic tundra or Australian desert). Can the navigational capacities of humans be fully explained by the cognitive map theory of the hippocampus, or is there more going on?

  Eichenbaum most readily describes navigation as what it is not. “I think navigation is not about Cartesian maps,” he offered. “It’s a story or memory problem.” The hippocampus is not so much about spatial memory, he said, as it is about “memory space.” Parsing this distinction is important. True navigation, in his opinion, is what happens when we travel to an unseen place. It requires planning a future (envisioning the place we want to go), calculating or remembering the route to get there (a sequence or narrative), and then orienting to ensure we are on the right track, often by comparing our memory (or perhaps a description we’ve been told) to our real-time perception of movement through space. “There are huge memory demands to solving the problem of navigation,” he said. “Memory steps in at every moment.”

  Space and its role in hippocampal function has been oversold, says Eichenbaum, for whom space is just one of many “fabrics” that we hold memory in. He believes that the hippocampal cells called place cells are much more flexible and capable of adapting to different dimensions. One of those dimensions is temporal, and for this reason, Eichenbaum doesn’t call them place cells, he calls them time cells. “Time is a philosophically interesting question. Do we make it up?” he mused. “As you navigate, you are moving in space and time together, and the hippocampus is mapping both.” His research has led him to believe that the organization of our episodic memories is supported by these time cells, and that mapping sequences of memories in time is just as critical to navigation as mapping geographic space. The trick is trying to design experiments that can demonstrate the difference because “you can’t usually parse space and time.”

  He motioned me to his desk and opened a video file on the computer. I was looking down at the faint outline of a healthy, plump white rat with black markings. Its head was obscured by wires attached to the electrodes inserted in its brain. Eichenbaum had conducted this particular experiment in the laboratory across the hall a few years earlier. It appeared to be like so many others. A rat is released in a figure-eight maze with a reward at the end. But this one had a treadmill at the stem of the maze. Before the rat could find its way to the reward it had to step on the treadmill, which was programmed to randomly speed up and slow down. As the rat started running in place at these different speeds, the electrodes in its brain recorded the firing of three different hippocampal cells, represented by a colored pixel on the screen. “Watch closely,” said Eichenbaum. “First you’ll see a blue dot, then a green, and last a pink dot.”

  As the rat started to run, I saw each cell fire in the order Eichenbaum described. I could tell that watching the video still thrilled him four years later. But what did these colored pixel-neurons prove? By holding behavior (running) and location (in place) constant, and randomizing the treadmill speed, Eichenbaum had decoupled the distance the rat traveled from the time it spent running and could track which neurons were mapping each variable. The results show that the hippocampus was encoding both time and distance simultaneously. Then, when the treadmill stopped and the rat continued through the maze, the very same neurons began to fire because they were encoding space. Experiments like these, in which hippocampal cells “map” multiple dimensions, are why Eichenbaum believes the hippocampus is capable not only of organizing physical space but of creating “temporally structured experiences into representations of moments in time.”

  After years of studying rats in mazes, Eichenbaum has come to understand the hippocampus as the “grand organizer” of the brain. “It’s organizing and integrating all these bits and pieces of information in a contextual framework,” he said. “It does create a map. I’m all for the cognitive map in the original sense that it’s a map where you put stuff to remember where they are in relationship to each other. That is a specific, limited, concrete sense of moving in geographic space and how did I get from here to there. The other sense is this abstract term, how did I navigate graduate school? What’s the path to the presidency? In human language, these are both legitimate. But which one is the hippocampus? Is it the specific one or the generic one? I think the hippocampus could mean to map things in time. And there are other spaces in addition to geometric space. It doesn’t have to be Euclidean or linear. That’s just a really good example of what the hippocampus does, but it has other functions.”

  In the last five years there has been more interest in designing tests to explore what those other spaces could be. A few years ago, a team of researchers in New York and Israel wondered whether the hippocampus could map social space: the relationship and interactions among individuals with different roles and levels of power. They asked individuals to participate in a role-playing game in which they moved to a new town and had to find a place to live and work, and they found that the hippocampus was activated during the tasks, indicating it’s a circuit that is important for “navigating” social relationships. Another study, authored by Sundeep Teki and others in 2012 and called “Navigating the Auditory Scene,” found that professional piano tuners had something in common with London’s taxi drivers: more hippocampal gray matter. The more years a person spent tuning pianos, the larger this part of their brain was. Sound, in their case, was the space that the hippocampus mapped. Different pitches and beat rates were landmarks, and routes were created from one previously tuned note to another. A study in the same journal two years earlier reported that musical training actually induces plasticity in the hippocampus. After just two semesters of training, the researchers saw evidence in fMRIs that music academy students’ hippocampi had enhanced responses to hearing sounds. Had their hippocampal cells become music cells?

  Eichenbaum thinks that results like these are perhaps more faithful to the original idea of the cognitive map described by Tolman back in 1948. A close reading of that now-historic paper reveals that he thought the cognitive map might be multidimensional, a tool capable of mapping multitudes of life experiences. And these new studies are also relevant to how we answer the question Eichenbaum first posed to me: What is navigation really? Insights into time cells, social space, and music highlight how complex human navigation in the brain is: not just a calculation based on reading a Cartesian map but an unfolding memory or a narrative sequence, human relationships, sensory experiences, personal history, or paths into the future. “The hippocampal system,” Eichenbaum once wrote, is “encoding events as a relational mapping of objects and actions within spatial contexts, representing routes as episodes defined by sequences of places traversed.”

  Sometimes that sequential story is geographic. Sometimes that episodic narrative is who we met and the words that were spoken. Sometimes that memory is a piece of music that carried us on a journey.

  * * *

  The notion of a map guiding
our movements is so deeply pervasive, a metaphor so cherished by the Western mind that it can feel nearly impossible to transcend. How could we know our way without a map? How can most of us, even children, sit down and draw maplike representations of familiar places if we don’t already possess one inside our heads?

  Throughout history, scientists have turned to material artifacts as metaphors for understanding the processes of life. Kepler described the universe as a clock. Descartes described reflexes as a push-and-pull system typical of sixteenth-century technology. Tolman’s contribution was to depart from the telephone switchboard metaphor to a map. Today it’s common to see the human brain likened to a computer, and the hippocampus to a GPS. Can these metaphors really capture the complexity of biology, or do we reach for them because we lack imagination for what’s really at work? “The cognitive map is a metaphor for what the brain does,” the neuroscientist Hugo Spiers told me, “but the problem with maps is they are complicated ideas in and of themselves. They are a type of metaphor already.”

  The philosopher William James called this problem the psychologist’s fallacy. James was concerned that scientists so often mistake the product of reflection and analysis of our experience as characteristic of immediate experience. Yet when we reflect and analyze, we are already stepping outside our direct experience in order to give an account of it, already beginning the process of reaching for metaphors that can only fall short in capturing our experience. Often those metaphors and models we reach for are influenced by human tools, not innate cognitive processes. James’s philosophy is called “radical empiricism,” and he believed that humans were capable of directly and objectively perceiving the world.